© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 24154
    Prev
    Next
    INDEX
    Explanations

    the word "who" and its contextual variations, indicating a focus on identifying individuals or questioning identity

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    320
    +0.10
    0.5%
    735
    +0.10
    0.5%
    679
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    126
    +0.10
    0.03
    692
    +0.10
    0.03
    735
    +0.10
    0.03
    Negative Logits
    framework
    -0.77
    MER
    -0.70
    emin
    -0.68
    strip
    -0.67
     Globe
    -0.67
    retion
    -0.67
     Glob
    -0.65
    warning
    -0.64
    emp
    -0.62
     Warning
    -0.62
    POSITIVE LOGITS
     else
    1.11
    soever
    1.06
     owns
    0.98
     exactly
    0.92
     cares
    0.92
     cared
    0.88
    abouts
    0.86
     participates
    0.85
     owes
    0.85
    dinand
    0.84
    Activations Density 0.050%

    No Known Activations